Interview with Ai2's Former RLHF Lead: Open Models Rise and RSI Myths
ziv_ravid · x · 2026-08-09
This podcast episode features Nathan Lambert, who previously led post-training at Ai2 and authored the book on RLHF. The discussion centers on the rapid improvement of open-source models and their implications for the next 2-5 years.
Lambert shares several key insights: he was surprised by how good open models have gotten recently; he believes the hype around recursive self-improvement (RSI) is currently overblown; and he predicts that Anthropic's famously open internal culture might crack under the pressures of an IPO.
More from Companies & People
- Tim O'Reilly: DeepMind Shakeup Isn't a Retreat, Google is Pivoting to Cloud — gaganghotra_ · 2026-08-09
- Former OpenAI Policy Advisor: It's Appropriate to Be Bummed About AI Right Now — Miles_Brundage · 2026-08-09
- Meme Mocks Big Tech Bureaucracy: Execs Debate Simple Feature for 3 Years — dreamwieber · 2026-08-09
- KPMG Survey: 49% of Companies Cut AI Agent Rollouts as Costs Outrun Value — danie-l · 2026-08-09
- Debate: Will Vertical AI Giant Harvey Achieve AGI Faster Than Anthropic? — Shahules786 · 2026-08-09
- 3 Weeks at xAI: Inside the Rocket-Speed Team Culture — wavefnx · 2026-08-09